Salmonella enterica is a facultative anaerobic gram-negative bacterium with more than 2,600 serotypes and an important zoonotic pathogen worldwide (Grimont and Weill, 2007). Human consumption of Salmonella-contaminated products can cause diarrhea, intestinal inflammation, and even bacteremia (Fearnley et al., 2011; Chousalkar et al., 2018; Tack et al., 2019), adding to the global burden of disease. Salmonella was previously reported to have caused approximately 22.2% (12,769 cases) of foodborne illness cases in China between 1994 and 2005 (Wang et al., 2007), and one of the sources of illness was waterfowl. China is the largest producer and consumer of waterfowl, such as ducks and geese, and related products (Wang et al., 2017). Recent studies have shown that waterfowl are an important source of Salmonella (Wang et al., 2020a; Kim et al., 2021) and often transmit Salmonella due to open-yard feeding (Murray et al., 2021); however, relevant systematic studies on Salmonella prevalence in these species have rarely been reported.

Because of diverse Salmonella serotypes and possible monophasic variation, simple slide agglutination assays to detect rare serotypes are often labor intensive and time consuming and have a risk of misidentification (Uelze et al., 2020). Sequence-based serotyping approaches, such as the Salmonella in silico typing resource (SISTR), can be used as a complementary method; this method was reported to have an accuracy of up to 94% (Yoshida et al., 2016). The in silicon method allows the detection of antigen genes carried by an isolate, while the slide agglutination method allows the detection of antigens expressed by an isolate (Yachison et al., 2017). A combination of these two methods can yield more accurate results. Pulsed-field gel electrophoresis (PFGE) has emerged as a method for analyzing large molecules of DNA (Sharma-Kuinkel et al., 2016) and has been widely used in molecular epidemiological investigations of foodborne pathogens (e.g., Escherichia coli, Salmonella enterica and Listeria monocytogenes; Favier et al., 2013; Li et al., 2020a). PFGE results reflect the genetic relationships among different isolates, allowing the rapid monitoring, tracking and tracing of bacterial infections.

In recent decades, antimicrobial agents have been used frequently in animal husbandry not only to treat and control Salmonella and other pathogens but also as prophylactic measures and growth-promotors. The abuse and misuse of antimicrobials has led to antimicrobial resistance (AMR), which is still increasing (Palma et al., 2020). To address this problem, use of antimicrobials as a growth-promoting factors has been banned in the European Union since 2006 (Castanon, 2007). China is one of the world’s largest producers and consumers of antimicrobials, with 162,000 tons of antimicrobials used in 2013, 52% of which were for veterinary use (Zhang et al., 2015). Over the past two decades, China’s restrictive policies on veterinary antimicrobials have changed substantially and become increasingly stringent (Yinqi et al., 2019). Consequently, the prevalence of colistin-resistant Escherichia coli in pigs and chickens decreased dramatically from 2015 to 2018 due to the withdrawal of colistin as an animal growth promoter in China (Wang et al., 2020b), highlighting the impact of addressing AMR. Nevertheless, there are still numerous reports showing that the Salmonella prevalence in poultry, pigs and eggs in China had exhibited increasing resistance to multiple antimicrobials (Yang et al., 2019; Li et al., 2020b, 2021; Xu et al., 2021). Thus, it is of interest and importance to perform continuous AMR monitoring and to investigate changes in AMR over time, which is crucial in identifying the mechanisms involved and providing guidance on rational treatment strategies. Characterization of Salmonella resistance is mainly performed by determining phenotypes based on antimicrobial susceptibility testing (AST; CLSI, 2020) and genotypes based on whole-genome sequencing (WGS; Schwan et al., 2021; Yan et al., 2021; Medina-Santana et al., 2022). Phenotypes usually correspond to the external expression of a single gene but may also be the result of synergistic effects of multiple genes (Morales et al., 2005). Therefore, both phenotypic and genotypic testing are necessary for pathogen surveillance and diagnosis, and using them together can lead to more accurate judgments.

In this study, 126 Salmonella isolates (including 65 collected in 2018–2020 and 61 collected in 2002–2005) obtained from waterfowl samples in Sichuan, China, were subjected to serotyping, PFGE molecular typing, AST, and WGS to investigate their prevalence and AMR profiles.

The seven serotypes identified in this study showed some diversity, and the serovars Enteritidis, Montevideo, Potsdam and Typhimurium are also reported to be the major serotypes prevalent in the global poultry industry (Shah et al., 2017; Yang et al., 2020; Diaz et al., 2021). The high homology of the PFGE patterns among 58 serovar Enteritidis isolates and 34 serovar Potsdam isolates indicated that a dominant clone was prevalent locally. In contrast, the remaining isolates showed quite dissimilar PFGE patterns, suggesting that they have genotypic diversity and that various Salmonella clones were prevalent in different waterfowl farms in Sichuan, China. As with many epidemiological surveys, the prevalence of a particular pathogen is usually regional in nature (Yan et al., 2021). Comparisons of PFGE and serotype results revealed the following three association patterns. First, most of the isolates with the same serotypes exhibited similar PFGE band distributions and thus were grouped into the same PFGE cluster, such as serovar Enteritidis and Potsdam isolates; this indicated that their genomic arrangements were similar and that their homology was relatively high. Second, several isolates with the same serotype, such as serovar Cerro, exhibited significant heterogeneity in their PFGE patterns, with alterations in the band distributions, suggesting that these isolates may have undergone genomic rearrangement, resulting in changes in the enzymatic cutting site that did not affect their surface antigen composition. This phenomenon is supported by a previous report that 46 Salmonella enterica serovar Schwarzengrund isolates, with a considerable length of evolutionary time, are still of the same serotype (Yang et al., 2022). Finally, individual isolates exhibited similar pulsotypes but different serotypes; this was observed between the serovar Potsdam isolate RCAD-S-015 (6,7,14:l,v,e,n,z₁₅) and the serovar Montevideo isolate RCAD-S-016 (6,7,14,[54]:g,m,s:-). Previous reports have also shown that epidemiologically unrelated isolates can be assigned to identical PFGE types (Barco et al., 2013; Shi et al., 2015), such as serovar Typhimurium (1,4,[5],12:i:1,2) versus 4,5,12:i:- (Ranieri Matthew et al., 2013) and serovar Thompson (6,7,14:k,1,5) versus 1,7:-:1,5 (Soyer et al., 2010). Since XbaI PFGE may not be discriminatory enough in some cases, we confirmed by WGS that the level of ANI between these two isolates was not as high as those between the same serotypes.

Owing to the excessive use of antimicrobial agents in animal husbandry over the last few decades and the horizontal spread of resistance genes, AMR in Salmonella has become a major concern (Foley et al., 2008). All isolates identified in this study were MDR and were broadly resistant to conventional antimicrobials, suggesting that most bacteria have acquired resistance traits under prolonged selection pressures, which is consistent with many previous findings (Yang et al., 2020; Sun et al., 2021; Wang et al., 2021). We also found variations in the resistance patterns of isolates from different periods, with the isolates obtained in the first collection period exhibiting a marked increase in resistance to EFT and newer-generation antimicrobials as well as an increase in susceptibility to TET. This trend may be attributed to the “strictest antibiotic regulation” enacted by the Chinese government in 2012, which allowed the use of veterinary antimicrobials such as EFT (Yinqi et al., 2019); increased resistance to EFT was also reported in Salmonella from pigs in Sichuan from 2009 (11.4%) to 2014 (53.8%; Xiuying et al., 2015). In contrast, a study from Canada showed the opposite trend, with resistance to EFT decreasing from 62 to 18% from 2004 to 2008 (Dutil et al., 2010). Variations in resistance over time suggest that the resistance phenotypes may change with the use of veterinary antimicrobials in upcoming years (Hornish and Kotarski, 2002; Sato et al., 2014). In particular, a small proportion of isolates also showed moderate resistance to human-restricted antimicrobials, such as IPM and SXT, implying possible cross-transmission between humans and animals or, more likely, the acquisition of new specific resistance genes (Wall et al., 2016).

WGS has become a reliable method for the detection of resistance genes, allowing the accurate identification of individual resistance genes in addition to the identification of single nucleotide mutations; this knowledge may even be applied to predict unknown resistance genes according to their conserved structural domains (Köser et al., 2014; Rokney et al., 2020). Specific resistance genes against different types of antimicrobials were detected in the 49 isolates. Some of the resistance genomic and phenotypic characteristics were consistent; for instance, the presence of the catII and tet genes conferred CHL and TET resistance phenotypes, respectively. However, there were substantial inconsistencies, and some isolates did not contain specific resistance genes but showed a resistance phenotype, such as serovar Potsdam isolated from goose eggs with the CIP phenotype, which may be attributed to unknown resistance mechanisms or nonspecific functions of multiple redundant efflux pump-like genes. As previously reported, CIP resistance is influenced by the coordination between multiple genes (gyrA, gyrB, parE, and acrB; O'Regan et al., 2009). Additionally, some isolates possessed specific resistance genes but showed susceptibility to the corresponding antimicrobial, which may be due to mutations or functional incompleteness of the gene; for example, bla_TEM-116 in RCAD-S-005 was not observed to be resistant to beta-lactams. Eight types of incompatible plasmid replicon fragments were detected in these 49 isolates. It is reasonable to assume that the AMR differences between the representative isolates RCAD-S-040 and RCAD-S-045 stems from the different plasmids they contain, and the same situation exists for RCAD-S-014 and RCAD-S-16. Since WGS does not provide a complete map, it is not possible to determine which plasmid a specific resistance gene is located on. As plasmids play a vital role in the horizontal transfer of resistance genes, their sequences and contributions to AMR need further investigation. Overall, the analysis of resistance phenotypes and genotypes suggests that their AMR profiles may be a result of long-term stress through mechanisms such as target alteration or horizontal gene transfer by mobile genetic elements (Bakkeren et al., 2022; Cohen et al., 2022).

Since our samples were obtained from waterfowl with pathological symptoms, we obtained a higher isolation rate than those reported in other studies and consequently a higher rate of AMR (Chen et al., 2020; Han et al., 2020). The shortcomings of our study include an insufficient sample size and the lack of a wide geographical distribution. Nevertheless, our study not only complements the epidemiological surveillance data for monitoring Salmonella of waterfowl origin but also has practical implications for guiding the use of antimicrobial agents in waterfowl in this region.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found at: NCBI: PRJNA857101.

XZ and AC conceived the study and edited the manuscript. YG and YL performed most of the research and drafted the manuscript. JL, DZ, RJ, ML, MW, and SC participated in sample collection and bacterial isolation. ZY analyzed the genome sequences. SZ, QY, YW, and QG did the PFGE detection. XO, SM, JH, DS, and BT participated in analysis of antimicrobial resistance. All authors contributed to the article and approved the submitted version.

This research was supported by the National Natural Science Foundation of China (32072877), the Applied Basic Research Programs of Science and Technology Department of Sichuan Province (2019YJ0436), and the earmarked fund for China Agriculture Research System (CARS-42-17).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.